def test_delay_zero_cancels_all(self):
delay = 0
X_hat = wpe.wpe(self.Y, self.K, delay=delay)
# Beginning is never zero. It is a copy of input signal.
tc.assert_allclose(X_hat[:, delay + self.K - 1:],
np.zeros_like(X_hat[:, delay + self.K - 1:]),
atol=1e-10)
def test_wpe_v1_vs_v7(self):
desired = wpe.wpe_v1(self.Y, self.K, self.delay)
actual = wpe.wpe_v7(self.Y, self.K, self.delay, mode='cut')
tc.assert_allclose(actual, desired)
actual = wpe.wpe_v7(self.Y, self.K, self.delay, mode='pad')
with tc.assert_raises(Exception):
tc.assert_allclose(actual, desired, rtol=1e-3, atol=1e-3)
def test_correlations_narrow_v1_vs_v5(self):
inverse_power = wpe.get_power_inverse(self.Y)
R_desired, r_desired = wpe.get_correlations_narrow(
self.Y, inverse_power, self.K, self.delay)
R_actual, r_actual = wpe.get_correlations_narrow_v5(
self.Y, inverse_power, self.K, self.delay)
tc.assert_allclose(R_actual, R_desired)
tc.assert_allclose(r_actual, r_desired)
def test_filter_operation_v1_vs_v4(self):
filter_matrix_conj = np.random.normal(size=(self.K, self.D, self.D)) \
+ 1j * np.random.normal(size=(self.K, self.D, self.D))
desired = wpe.perform_filter_operation(self.Y, filter_matrix_conj,
self.K, self.delay)
actual = wpe.perform_filter_operation_v4(self.Y, filter_matrix_conj,
self.K, self.delay)
tc.assert_allclose(desired, actual)
def test_correlations_v1_vs_v2_toy_example(self):
K = 3
delay = 1
Y = np.asarray([[11, 12, 13, 14], [41, 22, 23, 24]], dtype=np.float32)
inverse_power = wpe.get_power_inverse(Y)
R_desired, r_desired = wpe.get_correlations(Y, inverse_power, K, delay)
R_actual, r_actual = wpe.get_correlations_v2(Y, inverse_power, K,
delay)
tc.assert_allclose(R_actual, R_desired)
tc.assert_allclose(r_actual, r_desired)
def test_filter_matrix_conj_v1_vs_v5(self):
inverse_power = wpe.get_power_inverse(self.Y)
correlation_matrix, correlation_vector = wpe.get_correlations(
self.Y, inverse_power, self.K, self.delay)
desired = wpe.get_filter_matrix_conj(correlation_matrix,
correlation_vector, self.K,
self.D)
actual = wpe.get_filter_matrix_conj_v5(self.Y, inverse_power, self.K,
self.delay)
tc.assert_allclose(actual, desired, atol=1e-10)
def test_correlations_narrow_v1_vs_v6(self):
inverse_power = wpe.get_power_inverse(self.Y)
R_desired, r_desired = wpe.get_correlations_narrow(
self.Y, inverse_power, self.K, self.delay
)
s = [Ellipsis, slice(self.delay + self.K - 1, None)]
Y_tilde = wpe.build_y_tilde(self.Y, self.K, self.delay)
R_actual, r_actual = wpe.get_correlations_v6(
self.Y[s], Y_tilde[s], inverse_power[s]
)
tc.assert_allclose(R_actual.conj(), R_desired)
tc.assert_allclose(
r_actual.conj(),
np.swapaxes(r_desired, 1, 2).reshape(-1, r_desired.shape[-1]),
rtol=1e-5, atol=1e-5
)
def test_filter_matrix_conj_v1_vs_v7(self):
inverse_power = wpe.get_power_inverse(self.Y)
correlation_matrix, correlation_vector = wpe.get_correlations(
self.Y, inverse_power, self.K, self.delay
)
desired = wpe.get_filter_matrix_conj(
correlation_matrix, correlation_vector, self.K, self.D
)
s = [Ellipsis, slice(self.delay + self.K - 1, None)]
Y_tilde = wpe.build_y_tilde(self.Y, self.K, self.delay)
actual = wpe.get_filter_matrix_v7(
self.Y, Y_tilde=Y_tilde, inverse_power=inverse_power,
)
tc.assert_allclose(
actual.conj(),
np.swapaxes(desired, 1, 2).reshape(-1, desired.shape[-1]),
atol=1e-10
)
def test_wpe_v1_vs_v4(self):
desired = wpe.wpe_v1(self.Y, self.K, self.delay)
actual = wpe.wpe_v4(self.Y, self.K, self.delay)
tc.assert_allclose(actual, desired)
def test_wpe_v6_vs_sc(self):
from cbj.wpe._numpy import sc_wpe
desired = wpe.wpe_v6(self.Y, self.K, self.delay)
actual = sc_wpe(self.Y, self.K, self.delay, iterations=3)
tc.assert_allclose(actual, desired)
def test_wpe_v6_vs_v7(self):
desired = wpe.wpe_v6(self.Y, self.K, self.delay)
actual = wpe.wpe_v7(self.Y, self.K, self.delay)
tc.assert_allclose(actual, desired, atol=1e-10)
